proxmox-mcp
Minimal MCP server for Proxmox VE — 38 tools for nodes, VMs, LXC, storage, and snapshots. Read-only by default, single Docker image, multi-arch.
README
<!-- mcp-name: io.github.akmalovaa/proxmox-mcp -->
Proxmox MCP server
<p align="center"> <img src="docs/banner.png" alt="proxmox-mcp — MCP server for Proxmox VE" width="720"/> </p>
Simple Proxmox MCP
<p align="center"> <img src="docs/logo.png" alt="proxmox-mcp logo" width="140"/> </p>
MCP server for managing Proxmox VE
38 tools — nodes, QEMU VMs, LXC containers, storage, cluster, snapshots.
Why this one?
- One image, multi-arch —
docker run ghcr.io/akmalovaa/proxmox-mcp:latestand you're done - Just env vars — no config files, no database, no state
- Read-only by default — destructive ops are gated behind an explicit
PROXMOX_RISK_LEVEL - Tiny codebase — pure stdio MCP over Proxmoxer, no HTTP server, no auth layer, no extras
- Raw JSON out — no formatting, no emoji; LLM gets clean data
Quick start
Image: ghcr.io/akmalovaa/proxmox-mcp:latest (multi-arch: amd64 + arm64).
1. Export credentials in your shell profile (~/.zprofile, ~/.zshrc or ~/.bashrc):
# base environment:
export PROXMOX_HOST=192.168.1.100
export PROXMOX_USER=root@pam
export PROXMOX_PASSWORD=your-password
# or use token auth (recommended):
export PROXMOX_TOKEN_NAME=mcp
export PROXMOX_TOKEN_VALUE=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
# optional:
export PROXMOX_RISK_LEVEL=read
Reload: source ~/.zprofile (or restart the shell).
2. Add to ~/.claude/settings.json (Claude Code) or claude_desktop_config.json (Claude Desktop):
{
"mcpServers": {
"proxmox": {
"command": "docker",
"args": ["run", "-i", "--rm",
"-e", "PROXMOX_HOST",
"-e", "PROXMOX_USER",
"-e", "PROXMOX_PASSWORD",
"ghcr.io/akmalovaa/proxmox-mcp:latest"]
}
}
}
or token auth:
{
"mcpServers": {
"proxmox": {
"command": "docker",
"args": ["run", "-i", "--rm",
"-e", "PROXMOX_HOST",
"-e", "PROXMOX_USER",
"-e", "PROXMOX_TOKEN_NAME",
"-e", "PROXMOX_TOKEN_VALUE",
"ghcr.io/akmalovaa/proxmox-mcp:latest"]
}
}
}
docker run -e VAR without a value passes the host variable through — no secrets in the config file. Restart the client — 38 Proxmox tools become available.
For password auth, swap the token vars for PROXMOX_PASSWORD.
Note: Claude Desktop on macOS is launched via launchd and does not inherit
~/.zprofile/~/.zshrc. Either put the exports in~/.zshenv, or fall back to an inline"env": { ... }block in the config.
Configuration
All settings are environment variables — set them in your shell profile, pass them inline to docker run -e, or declare them in your MCP client's env block.
| Variable | Default | Description |
|---|---|---|
PROXMOX_HOST |
— | Proxmox host (IP or hostname) |
PROXMOX_USER |
root@pam |
API user |
| Auth | — | token or password — see below |
PROXMOX_PORT |
8006 |
API port |
PROXMOX_VERIFY_SSL |
false |
Verify TLS certificate |
PROXMOX_RISK_LEVEL |
read |
read / lifecycle / all |
Authentication: token or password
Pick one. If both are set, the token wins.
Token (recommended) — create in Proxmox UI: Datacenter → Permissions → API Tokens → Add (uncheck Privilege Separation). Then:
export PROXMOX_TOKEN_NAME=mcp
export PROXMOX_TOKEN_VALUE=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
Password (fallback):
export PROXMOX_PASSWORD=your-password
Risk levels
PROXMOX_RISK_LEVEL gates destructive operations:
| Level | Adds |
|---|---|
read (default) |
read-only tools |
lifecycle |
+ start / stop / reboot / suspend / clone / create-snapshot |
all |
+ delete-snapshot / rollback-snapshot |
Every elevated call is logged to stderr (ALLOW / DENY + tool + tier).
Tools
Nodes (7)
| Tool | Description |
|---|---|
list_nodes |
List all cluster nodes with status, CPU, memory, uptime |
get_node_status |
Detailed node metrics (CPU, memory, disk, load, kernel) |
get_node_networks |
Network interfaces on a node |
get_node_disks |
Physical disks on a node |
get_node_tasks |
Recent tasks on a node |
get_task_status |
Status of a specific task by UPID |
get_task_log |
Log output from a task |
QEMU VMs (14)
| Tool | Tier | Description |
|---|---|---|
list_vms |
read | List all VMs, optionally filter by node |
get_vm_status |
read | Current VM status (running/stopped, CPU, memory) |
get_vm_config |
read | VM configuration (hardware, disks, network) |
list_vm_snapshots |
read | List all snapshots of a VM |
start_vm |
lifecycle | Start a VM |
stop_vm |
lifecycle | Force-stop a VM |
shutdown_vm |
lifecycle | Graceful ACPI shutdown with timeout |
reboot_vm |
lifecycle | Reboot via ACPI |
suspend_vm |
lifecycle | Suspend a VM |
resume_vm |
lifecycle | Resume a suspended VM |
clone_vm |
lifecycle | Full or linked clone |
create_vm_snapshot |
lifecycle | Create a snapshot |
delete_vm_snapshot |
all | Delete a snapshot |
rollback_vm_snapshot |
all | Rollback to a snapshot |
LXC Containers (11)
| Tool | Tier | Description |
|---|---|---|
list_containers |
read | List all LXC containers, optionally filter by node |
get_container_status |
read | Current container status |
get_container_config |
read | Container configuration |
list_container_snapshots |
read | List all snapshots |
start_container |
lifecycle | Start a container |
stop_container |
lifecycle | Force-stop a container |
shutdown_container |
lifecycle | Graceful shutdown with timeout |
reboot_container |
lifecycle | Reboot a container |
create_container_snapshot |
lifecycle | Create a snapshot |
delete_container_snapshot |
all | Delete a snapshot |
rollback_container_snapshot |
all | Rollback to a snapshot |
Storage (2)
| Tool | Description |
|---|---|
list_storage |
Storage pools with usage, optionally filter by node |
get_storage_content |
Contents of a storage pool (ISOs, backups, images, templates) |
Cluster (4)
| Tool | Description |
|---|---|
get_cluster_status |
Cluster health, quorum, node membership |
get_cluster_resources |
All resources (VMs, containers, storage, nodes) |
get_cluster_backups |
Configured backup jobs |
get_next_vmid |
Next available VM/container ID |
Architecture
src/proxmox_mcp/
├── server.py # FastMCP instance + entry point
├── config.py # Pydantic Settings (PROXMOX_ prefix)
├── client.py # Proxmoxer connection via lifespan
└── tools/ # nodes, vms, containers, storage, cluster
- Read-only by default — elevated tools gated by
PROXMOX_RISK_LEVEL - Single connection — Proxmoxer client created once at startup, shared via lifespan
- Raw JSON output — no formatting; LLM consumes data directly
Development
Run standalone (testing)
export PROXMOX_HOST=192.168.1.100
export PROXMOX_USER=root@pam
export PROXMOX_TOKEN_NAME=mcp
export PROXMOX_TOKEN_VALUE=xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx
docker run -i --rm \
-e PROXMOX_HOST -e PROXMOX_USER \
-e PROXMOX_TOKEN_NAME -e PROXMOX_TOKEN_VALUE \
ghcr.io/akmalovaa/proxmox-mcp:latest
Without Docker (UV)
git clone https://github.com/akmalovaa/proxmox-mcp.git && cd proxmox-mcp && uv sync
MCP client config:
{
"mcpServers": {
"proxmox": {
"command": "uv",
"args": ["run", "--directory", "/path/to/proxmox-mcp", "python", "-m", "proxmox_mcp"],
"env": {
"PROXMOX_HOST": "192.168.1.100",
"PROXMOX_TOKEN_NAME": "mcp",
"PROXMOX_TOKEN_VALUE": "xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
}
}
}
}
Build from source
git clone https://github.com/akmalovaa/proxmox-mcp.git
cd proxmox-mcp
docker build -t proxmox-mcp .
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。